V-Nova – Sustainable streaming: the state of play and what’s next
Sam-Orton Jay, VP Product, V-Nova
Sustainability is undeniably a pressing concern within the video streaming industry, and the latest data about emissions generated by the sector underscores the urgency of addressing its environmental impact. As has been widely quoted, with between 2% and 4% of global energy usage accounted for by ICT and with more than 70% of internet traffic associated with video, it is clear that improving our energy footprint can have a significant impact on the problem overall.
However, as we navigate the sustainability landscape, it is essential to avoid succumbing to overly simplistic explanations or superficial solutions that might inadvertently become greenwashing. The complexity of the issue demands nuanced, well-researched strategies that genuinely contribute to a greener future. Delving deeper into the sustainability of video streaming reveals the inherent challenge of quantifying and estimating the impact of its workflow components accurately. Remarkably, there remains a lack of consensus across the industry regarding what metrics to measure and how to measure them, let alone how to mitigate these impacts effectively.
The present state of play
More and more companies are beginning to evaluate the energy usage or carbon footprint of their products and solutions. This is driven by a growing realization that there are some win-win opportunities out there where reducing energy consumption is both good for the environment and for the business’ bottom line. The limitations of these initiatives though, are often that it’s difficult for companies to get a holistic view of their part in a much wider ecosystem. This is critical to ensure that changes do ultimately have the positive effect they’re intended to and don’t trigger unintended consequences.
It was for this reason that at V-Nova we were keen to fully understand the sustainability implications of the use of one of our own technologies, namely MPEG-5 LCEVC. We already knew that LCEVC can deliver higher-quality video at up to 40% lower bitrates, reduce transcoding energy usage by up to 70% and extend the life cycle of existing devices by enhancing the capabilities of the native codecs they support. But which of these would have the most positive impact on energy usage and in turn, carbon emissions overall? And would there be any potential effects on energy consumption in the rest of the systems that we play a part in?
Challenges preventing the industry from doing more
The single biggest challenge to us collectively reducing energy consumption is the lack of data on where the issues really lie, and that is in turn because of how difficult it has become to measure what are very often long workflows with many components.
Most of the research that has been drawn on to date is based on quite limited data sets, which is meaning that the validity of research like the Shift Project which mapped emissions directly to each GB of internet traffic is being increasingly questioned. This is fundamentally because as an industry we are at a very early stage when it comes to measuring power consumption across workflows. For example, determining how much power your video player is consuming on a mobile device is very challenging because the dominant operating systems weren’t designed to measure or report this. Similarly, if you’re running video encoding in public cloud infrastructure, where do you start to determine the power consumption when your virtual machine could be sharing hardware (or not) with a multitude of other processes?
When the research isn’t covering things holistically enough and is forced to make large assumptions due to a lack of data, there is a danger that we can make changes which either have quite negligible benefits or could even make things worse.
One possible approach where the benefits are unclear is to reduce the amount of data we deliver either by changing encoding strategies or by adjusting players or backend systems to favor lower bitrate profiles. It’s perfectly logical to think that reducing the data delivered means reducing energy usage. However, it is hard to fully quantify the impact of, say, a 20% reduction in bandwidth. Networks and most of our backend infrastructure like CDNs are provisioned for peak traffic and are largely “always on”. Lowering overall streaming traffic will help, but the actual overall impact is something we must measure, as it is certainly not a linear relationship as some believe.
These challenges all point to the need for industry-wide collaboration between vendors and services to start measuring our impact in detail and to figure out how we make a significant difference. Thankfully, some such efforts are beginning…
Pragmatic steps toward sustainability
We should not be overwhelmed and paralyzed by the complexity of the issue as there are tangible steps we can take to begin our journey towards sustainability. At V-Nova we’ve found it tremendously valuable to be working alongside the experts in Greening of Streaming on trying to understand, measure, experiment and ultimately find approaches to significantly reduce our collective impact through the Low Energy Sustainable Streaming (LESS) Accord. The great projects we are embarking on in that group to test a variety of approaches and provide the best practices recommendations that will form the LESS Accord should give us all reliable guidance to follow. Until then, here are a few concrete actions we can implement that will start us in the right direction:
Start measuring and targeting
Challenge your own teams to work towards establishing meaningful KPIs around energy usage. That will almost certainly mean that the first step must be initiating a lot more measurement to understand what the current energy usage is and where the biggest problems in your systems lie.
Optimize system utilization
We must explore ways to minimize the utilization of existing systems without adversely affecting other operations. For instance, services could evaluate whether retaining everything in storage or caches is truly necessary or if infrastructure can be pruned to reduce emissions.
Reduce redundant workflows
Addressing duplicated workflows, especially for various device categories, is crucial. These redundancies often lead to a significant increase in the energy consumption required for transcoding and content delivery. Technologies such as MPEG-5 LCEVC, which has seen extensive trials by Globo in Brazil, can offer a path to enhance the quality of user experiences while optimizing existing pipelines and devices.
Apply upgrades with energy efficiency in mind
When considering hardware upgrades, prioritize appliances that consume less power. The choice of equipment can substantially impact the overall energy efficiency of streaming operations. However, factor in the embedded emissions of new equipment when evaluating whether to upgrade. Implementing technologies such as LCEVC can extend the life cycle of existing consumer devices by enhancing the capabilities of the native codecs they support.
In conclusion, addressing the sustainability challenges of video streaming requires a multifaceted approach that combines technological innovation, industry collaboration, and a commitment to long-term environmental responsibility. By heeding these strategies and collectively working towards sustainability, the video streaming sector can take significant strides in reducing its ecological footprint.